基于DWT-PCA-LIBSVM的电能质量扰动分类方法  

Power Quality Disturbance Classification Based on DWT-PCA-LIBSVM

作　　者：李家俊 吴建军 陈武 钟建伟 LI Jia-jun;WU Jian-jun;CHEN Wu;ZHONG Jian-wei(Enshi Power Supply Company,State Grid Hubei Electric Power Co.,Ltd,Enshi 445000,China;College of Intelligent Systems Science and Engineering,Hubei Minzu University,Enshi 445000,China)

机构地区：[1]国网湖北省电力有限公司恩施供电公司,湖北恩施445000 [2]湖北民族大学智能科学与工程学院,湖北恩施445000

出　　处：《电工电气》2023年第3期20-24,共5页Electrotechnics Electric

基　　金：国家自然科学基金项目(62163013);国网湖北省电力有限公司2022科技项目(5215P0220001)。

摘　　要：针对传统电能质量扰动分类方法中特征混叠和分类精确度低等问题,提出一种基于DWTPCA-LIBSVM的扰动分类方法。对于常见的9种电能质量扰动信号,利用离散小波变换(DWT)提取不同扰动信号的特征向量,并将其按比例划分成训练集和测试集;采用主成分分析(PCA)方法将训练集和测试集数据降维处理;基于LIBSVM工具箱构建电能质量扰动分类模型进行分类识别。仿真实验结果表明:该方法能有效识别典型的9种电能质量扰动信号(包括两种复合扰动),验证了该方法对电能质量扰动信号分类的有效性。The traditional power quality disturbance classification has disadvantages of overlapping characteristics and low classification accuracy,so a new disturbance classification method based on DWT-PCA-LIBSVM is proposed.The paper adopts Discrete Wavelet Transform(DWT)to extract feature vectors of nine common power quality disturbance signals,and divides them into training sets and testing sets in proportion.Then,Principal Component Analysis(PCA)is used to reduce data dimension of training sets and testing sets,and power quality disturbance classification model is built to classify and recognize signals based on LIBSVM toolbox.The simulation results show that this method can efficiently identify nine typical power quality disturbance signals(including two composite disturbances),and verify its effectiveness of classification.

关 键 词：电能质量 离散小波变换 主成分分析 支持向量机 

分 类 号：TM60[电气工程—电力系统及自动化]